Controlling indecisive swarms

Controlling swarms of multi-agent systems using external agents remains a fundamental challenge across domains, from biology to swarm robotics. The key difficulty lies in managing agent interactions, as simple local interactions can give rise to complex global dynamics. In this talk, I will introduce a stochastic interaction algorithm called the Indecisive Swarm Algorithm (ISA), inspired by the behavior of small flocks of indecisive sheep in the presence of a shepherd dog. Unlike traditional approaches where agents follow leaders or average behaviors of other agents, ISA agents follow one agent or controller at a time and switch indecisively between them. This makes the ISA swarm a stochastic, non-reciprocal temporal network. I will demonstrate that deliberate indecisiveness can enhance swarm controllability. These insights have broad applications in controlling complex networks such as heterogeneous robot swarms, ecological networks, biochemical reaction networks, and social dynamics.